Loading page…
Meta says the model runs entirely on two M5 Ultra Mac Studios. Its downloadable weights allow customization, but commercial use remains subject to license conditions.
Listen to this story
Meta’s October 15 release makes Muse Spark Open downloadable for coding and sustained, multi-step work, with local deployment as an alternative to hosted inference for teams equipped with two M5 Ultra Mac Studios. The model has 816 billion total parameters, 42 billion active parameters and a one-million-token context; Meta says the paired Macs can hold it in unified memory. Commercial use is subject to a limited Muse Model License, so open weights do not mean unrestricted use, and Meta’s full evaluation report has not yet been published.
The desktop configuration uses Apple’s MLX framework and up to 512 GB of unified memory per Mac; Meta cites 1.2 TB/s memory bandwidth.
Meta demonstrated Muse Code completing Ice Cubes bug fixes, feature work and a full build, as well as using DaVinci Resolve to select and crop footage.
Helion kernels in SGLang showed 1.03–3.95× kernel speedups on NVIDIA and AMD GPUs; end-to-end improvement was 3% on AMD MI350, not the Mac setup.
Meta released Muse Spark Open’s model weights on October 15, 2026, giving developers a downloadable model built for coding and multi-step work. The company says it can run entirely on two Apple M5 Ultra Mac Studios—a local alternative to hosted access—though the release comes with limited commercial terms.
Muse Spark Open has 816 billion total parameters, with 42 billion active. Its context window—the material it can work with in a request—holds up to one million tokens. Meta positions it for complex software engineering, multi-step reasoning and tasks that require sustained work rather than a single answer.
Meta optimized the model for M5 Ultra Mac Studios using the latest MLX framework. Each machine’s integrated GPU can use up to 512 GB of unified memory. Together, Meta says, the two machines can hold the entire model. The company also cites the chip’s 1.2 TB/s memory bandwidth as sufficient to read the portion needed for each token.
In on-device tests through Muse Code, Meta says the model completed bug fixes, new features and a full build for Ice Cubes, an open-source Mastodon client. Another demonstration involved operating DaVinci Resolve, reviewing footage frame by frame to select moments and crop them vertically.
Separate GPU optimizations ship as automatically tunable Helion kernels in SGLang, software used to run the model. Meta’s performance figures below apply to NVIDIA and AMD hardware, not the desktop configuration.
Meta reports gains for individual Helion kernels on NVIDIA and AMD GPUs.
Meta reports this whole-system improvement on AMD MI350.
Hosted access offers a route that does not require setting up the desktop configuration. Meta lists these distribution channels:
Developers can further tune the model using PyTorch’s TorchTitan. The weights use the Muse Model License with limited commercial terms, which Meta says allow many developers to use the model for free.
Meta says it added safety post-training for the open-weight release and that the model met release thresholds under its Meta Superintelligence Scaling Framework. The company points to its model card for detailed results, with a full evaluation report promised soon.
Story updates
Meta released Muse Spark Open on October 15, 2026, with Together AI offering a brief period of free API use. But free requests do not mean a no-cost signup: new accounts must buy at least $15 in credits. The promotion also switches automatically to paid billing when it ends, rather than stopping requests.
Together AI’s offer runs through October 18, 2026, at 12:00 PM PT and applies only to Muse Spark Open on its Serverless API. Standard rate limits still apply, and Together reserves the right to change or end the offer early.
New accounts must add a payment method as well as purchase the credits required for activation. Together says those credits are not consumed while using the free model. After the promotion, requests are charged to the account’s payment method or credit balance at the following rates.
Meta describes Muse Spark Open as a natively multimodal model built for coding and agentic work—tasks that involve taking actions across multiple steps. It lists 816 billion total parameters, with 42 billion active, and a one-million-token context window for the material the model can work with in a request.
Together says training focused on real developer workflows, including changes spanning multiple files, debugging sessions and long-running assignments. Its description emphasizes tool calling, self-correction and coordinating subagents that divide up work. These are the provider’s capability claims, not independently verified results presented alongside the listing.
The hosted model accepts text, images, audio and video, while producing text output. Together also says it emits separate commentary explaining its intent before acting on longer tasks. Downstream applications can surface that commentary to show users what the model is doing and why.
For developers using Together’s Python client, the published example calls the promotional endpoint directly. Authentication uses a Together AI API key; the model identifier below is the one shown in its Python example.
Hosted access is only one route into the release. Meta’s launch announcement says the weights are available for download through Hugging Face and several cloud platforms. Other API providers include Baseten, Fireworks AI, Groq, Modal, Nebius and OpenRouter. Developers can also customize the model using PyTorch rather than rely solely on a hosted endpoint.
That access is not unrestricted. Meta is distributing the weights under the Muse Model License with limited commercial terms, which it says allow many developers to use the model for free. The license conditions are separate from Together’s temporary offer: one governs use of the released weights, while the other sets the hosted service’s promotional billing terms.
The performance evidence needs a separate reading. Together’s listing does not supply benchmark figures or full architecture details, saying those arrive with launch materials. Meta calls Muse Spark Open its most capable US open-weight model to date, but that ranking remains a company claim rather than an independent comparison established by the API listing.
Meta also says it added safety post-training specifically for the open-weight release and evaluated severe risks under its Superintelligence Scaling Framework. The company says the model met its release thresholds. It points to the model card for detailed results and says a full evaluation report will follow.
meta.aiLoading discussion...
Join the conversation
Explain what would make local operation worthwhile—or not—for you.
Be the first to share a perspective or an experience.
Reader comments
Newest comments first. Replies stay oldest first.